SmophyAI

Published daily · Next update 02:00 UTC

Google · Gemini family

Google: Gemini 3.7 Flash (batch)

Google: Gemini 3.7 Flash (batch) ranks #14 of 101 models by Artificial Analysis quality score (56) at $0.75/1M tokens blended - and ranks #12 of 312 by average real OpenRouter usage share over the trailing 54-day window (2.42%). Available from 2 providers at up to 100.0% uptime.

  1. Quality (Intelligence Index) - 56
  2. Coding index - 76.1
  3. Agentic index - 45.1
  4. Price (blended $/1M) - 0.75
  5. Context window - 1,048,576
  6. Usage rank - #12 of 312
  7. Providers - 2
  8. Open/closed - closed

Usage share over time

6.90%5.18%3.45%1.73%0.00%Share of usageAug 14Aug 20Aug 27

What Google: Gemini 3.7 Flash (batch) is actually used for

Share of each real task category's classified traffic this model handles, from OpenRouter's task classification.

TaskShare of task traffic
Shell Execution13.9%
Debugging5.4%
Workflow Execution4.7%
Math3.3%
Repo Scanning3.0%
Multi-step Planning2.8%
Tool Dispatch2.5%
Code Generation2.3%

Compare Google: Gemini 3.7 Flash (batch)

Frequently asked questions

What is Google: Gemini 3.7 Flash (batch)'s quality score?

Google: Gemini 3.7 Flash (batch) scores 56 on Artificial Analysis's Intelligence Index, ranking #14 of 101 models with a published score.

How much does Google: Gemini 3.7 Flash (batch) cost?

Google: Gemini 3.7 Flash (batch) costs $0.75 per 1M tokens (blended 75% prompt / 25% completion pricing).

Is Google: Gemini 3.7 Flash (batch) widely used?

Google: Gemini 3.7 Flash (batch) ranks #12 of 312 models by average real OpenRouter usage share, averaging 2.42% of classified token volume over the trailing 54-day window.

Is Google: Gemini 3.7 Flash (batch) open-weight or closed?

Google: Gemini 3.7 Flash (batch) is a closed-weight model.

Source: OpenRouter (openrouter.ai/rankings), as of August 28, 2026.

Benchmark scores: Artificial Analysis (artificialanalysis.ai) via OpenRouter (openrouter.ai/rankings).

Methodology: www.smophy.ai/benchmark/methodology

Token counts originate from each provider's own tokenizer and are not directly comparable across providers.